Zabbix caches some SNMP properties (engine time, engine boots, engine id) of a host when using SNMPv3 to enable RFC complaint communication with these hosts.
Unfortunately, we seem to be running into some snmpEngineTime issues on some devices every once in a while (we've double checked to see there are no duplicate SNMPengineID's in this environment).
The only current remedy we have if these issues occur, is to either reboot the device (which resets the engine time on the device and increments the engine boots counter), or we need to restart the Zabbix server proces (which clears the cache and is less of a problem normally).
It would be great if we could flush the snmp cache for these problem devices (via GUI and API) or if we could clear all of the SNMP cache with a -R flag to zabbix_server and zabbix_proxy without having to restart the entire server.
Relates to (amongst others):